Alain Martel <kua### [at] videotronca> wrote:
......
> adaptive 0 start with 4 samples (2 by 2): One at each corners.
> adaptive 1 start with 9 samples (3 by 3): Each corners plus one in the
> middle of each side and the center.
> adaptive 2 start with 25 samples (5 by 5), adaptive 3 with 81 (9 by 9),...
>
> This allow you to use something like this :
>
> area_light 3*x 3*z 1025 1025 adaptive 0
>
> and still have reasonably quick renders with no discernible banding,
> even without jitter.
>
> If you use an area_light of 4x4 with jitter and antialiasing, then the
> banding will disappears but your render time will greatly increase
> because all pixels in the penumbra will get antialiased.

code for chessdiff2 ---
+AM2 +A0.3
board_sq(<-3+j,0,-3+i>),so dimensions are: 3 * 8 = 24 and the frame.
#if (Area_Light)
  area_light x*15, y*15,6,6
  adaptive 1 jitter orient circular
#end
32 shadows takes 21 hrs. yikes!

during which time I surmised 4,4 would be enough, but it takes so long to check
these things. but 1025x1025, I cant even multiply that in my head, unbelievable
N lights = 1050625 = a million lights? seriously?

in chessdiff.png 13,13 - u can see the shadow streaks, which disappeared in
chessdiff2.png

so your saying, use a million lights, but adaptive 0 ?

Post a reply to this message